O R I G I N A L P A P E R
Doris HerrmannÆ Beat BollerÆBruno Studer Franco WidmerÆRoland Ko¨lliker
QTL analysis of seed yield components in red clover ( Trifolium pratense L.)
Received: 1 September 2005 / Accepted: 13 November 2005 / Published online: 6 December 2005 Springer-Verlag 2005
AbstractCultivars of red clover (Trifolium pratenseL.), an important forage crop in temperate regions, are often characterised by an unsatisfactory level of seed yield, leading to high production costs. This complex trait is influenced by many components and negatively correlated with other important traits, such as forage yield or persistence. Therefore, seed yield has proven to be difficult to improve. Thus, the objectives of this study were to assess association among seed yield components and to provide the basis for identifying molecular markers linked to QTLs for seed yield components to assist breeding for improved red clover cultivars. A total of 42 SSR and 216 AFLP loci were used to construct a molecular linkage map with a total map length of 444.2 cM and an average distance be- tween loci of 1.7 cM. A total of 38 QTLs were iden- tified for eight seed yield components. The traits seed number per plant, seed yield per head, seed number per head, head number per plant and percent seed set were highly correlated with seed yield per plant, and QTLs for several of these traits were often detected in the same genome region. Head number per plant may present a particularly useful character for the improvement of seed yield since it can easily be determined before seed maturity. In addition, two genome regions containing four or five QTLs for dif- ferent seed yield components, respectively, were iden- tified representing candidate regions for further characterisation of QTLs. This study revealed several key components which may facilitate further improve- ment of seed yield. The QTLs identified represent an
important first step towards marker-assisted breeding in red clover.
Introduction
Temperate grasslands play an important role in agri- culture, as they cover approximately 8% of the global land area (White et al. 2000). Red clover (Trifolium pratense L.), an outcrossing and diploid (2n=2x=14) species with a gametophytic self-incompatibility system, is an important component of permanent pastures and meadows as well as of grass-clover leys in temperate regions. Red clover is adapted to a wide range of envi- ronmental conditions, has a high nutritive value and, due to its ability to fix atmospheric nitrogen, red clover is of a high value to the environment (Taylor and Que- senberry1996).
In the last decades, targeted selection has produced red clover cultivars that are considerably improved for traits such as forage yield and quality as well as resis- tance and persistence (Taylor and Quesenberry 1996).
However, these improved cultivars often show an unsatisfactory seed yield that leads to high production costs and limits the success of cultivars in the market- place (Taylor and Quesenberry1996). Seed yield may be increased to a certain extent through improved man- agement practices involving irrigation (Oliva et al.1994) or selection of soil type (Belzile 1991). However, path coefficient analyses have revealed a causal relationship among several components associated with seed yield, highlighting the complexity of this trait (Montardo et al.
2003; Oliva et al. 1994). Although both studies consis- tently reported a significant effect of the number of heads per plant on seed yield, they were limited to the phenotypic observation of only a few seed yield com- ponents.
The negative correlation of seed yield with other important agronomic traits, such as forage yield (Steiner et al. 1997) represents a further impediment for
Communicated by T. Lu¨bberstedt
D. HerrmannÆB. BollerÆB. StuderÆF. WidmerÆR. Ko¨lliker (&) Agroscope FAL Reckenholz, Swiss Federal Research Station for Agroecology and Agriculture, 8046 Zurich, Switzerland E-mail: roland.koelliker@fal.admin.ch
Tel.: +41-44-3777345 Fax: +41-44-3777201
DOI 10.1007/s00122-005-0158-1
source: https://doi.org/10.24451/arbor.15238 | downloaded: 14.2.2022
improving this trait. This negative correlation seems to be particularly pronounced for persistence, i.e. the ability to produce constantly high forage yield across several growing periods. For example, continuous selection from locally adapted Swiss ecotypes led to the development of cultivars with considerably improved persistence, which are commonly referred to as Mattenklee cultivars (Herrmann et al. 2005). These cultivars show signifi- cantly increased forage yield across three or four growing periods when compared to other red clover cultivars (Lehmann and Briner1998). However, the seed yield of Mattenklee cultivars is considerably lower when com- pared to less persistent cultivars (Deneufbourg2004).
Molecular dissection of complex traits and the development of molecular markers linked to genes and quantitative trait loci (QTL) controlling such traits may provide new tools for breeding, which can complement traditional breeding approaches (Newbury2003). Iden- tification and integration of QTLs in genetic linkage maps is a promising step towards the development of molecular markers for marker-assisted breeding. Several examples have been reported for forage crops. In white clover (Trifolium repens L.) QTLs for seed yield and other seed yield components were recently identified (Barrett et al. 2005; Abberton et al. 2005). In perennial ryegrass (Lolium perenneL.) a genome region associated with four herbage quality traits was located in the vicinity of genes involved in lignin biosynthesis and is therefore a candidate region for more detailed charac- terisation of QTLs controlling herbage quality (Cogan et al. 2005). In major crops, such as soybean (Glycine maxL.) QTLs have been identified for a large number of traits including seed yield (Mansur et al. 1996), and in addition, molecular markers have been developed for application in breeding, for example to select for a soybean cyst nematode resistance allele (Cregan et al.
1999b). However, forage crops in general and red clover in particular have lagged behind in this rapid develop- ment. Currently, two linkage maps, a basic prerequisite for identification of markers linked to important traits, have been reported for red clover; one consisting of 157 RFLP markers (Isobe et al. 2003) and the other con- sisting of 1,357 SSR and 148 RFLP markers (Sato et al., submitted). However, so far there is no information available on QTLs controlling important traits, such as seed yield for red clover.
Moreover, there is only limited information available for association among seed yield components, restricted to phenotypic observations of only few seed yield com- ponents. The analysis of additional traits, such as seed yield per head or time of flowering as well as the molecular dissection of these traits may help to elucidate the association of seed yield components and to develop new strategies for seed yield improvement.
The objectives of the study presented here were: (1) to characterise the association among eight seed yield components, (2) to identify components which are easy to score and thus allow for improved selection and (3) to identify genome regions containing QTLs for seed yield
components for the future development of molecular markers for marker-assisted improvement of seed yield in red clover. For this purpose we established a red clover F1population segregating for seed yield compo- nents, investigated eight seed yield components in a field experiment, constructed a genetic linkage map using AFLP and SSR markers and performed QTL analysis.
Materials and methods
Plant material
A two-way pseudo-testcross population was created performing manual reciprocal pollinations and using two red clover genotypes with contrasting levels of seed yield. One genotype was selected from the cultivar Vi- oletta (pV), a Belgian cultivar characterised by a high seed yield but displaying low persistence (CLO-DvP, Ghent, Belgium). The other genotype was selected from the Swiss Mattenklee cultivar Corvus (pC), which is characterised by excellent persistence but rather shows low seed yield (Agroscope FAL Reckenholz, Zurich, Switzerland).
Seeds were harvested from each maternal plant sep- arately and a total of 400 seeds (200 per maternal plant) were germinated on wet filter paper (Schleicher and Schuell, Dassel, Germany) and transferred to soil-filled pots. Plants were grown for 3 months under long-day conditions [16 h light (‡100lEm 2s 1)] in the green- house. To obtain plants with as many shoots as possible, they were then cultivated for 5 months under short-day conditions [9 h light (275lEm 2s 1)] in the growth chamber. Clonal replicates were produced by cutting the plant in at least five parts, each containing equally sized longitudinal sections of the taproot. Roots were dipped in synthetic auxin (3–Indol Butyric Acid, Pokon and Chrysal International, Naarden, Netherlands) and re- grown for 4 months in the greenhouse under long-day conditions as described previously.
Experimental conditions and phenotypic evaluation In spring 2003, four clonal replicates of 280 genotypes (each parental plant serving as maternal genotype for a subset of 140 progenies) were planted in the field in a 4·4 lattice with the blocks arranged as a Latin square design, i.e. each genotype was represented once in each of the four rows and columns of the lattice, respectively.
The field experiment was established at Agroscope FAL Reckenholz in Zurich, where temperature and rainfall average 8.5C and 1,042 mm year 1, respec- tively. Plants were established in a clay soil together with Poa pratensiscv. Compact sown at the time of planting.
Phenotypic evaluation of the seed yield components was performed after the first cut in summer 2004. Eight traits were investigated: Seed yield per plant (g; SYP), seed number per plant (SNP), seed yield per head
(g; SYH), seed number per head (SNH), head number per plant (HNP), thousand-seed weight (g; TSW), percent seed set (number of seeds per 100 florets; PSS) and time of flowering (days after first cut; TOF). TOF was defined as the day when three heads of a plant were flowering. On maturity, one head per plant was used to count florets and seeds in order to calculate PSS. Seed yield and number of seeds of 20 randomly picked heads, as well as seed yield of the remaining heads were determined to calculate SYP, SNP, SYH, SNH, HNP and TSW.
Analysis of variance was performed using the general linear model (GLM) of the STATISTICA software (version 6.1, StatSoft, Tulsa, OK, USA). Least square means were used for all further calculations. Heritability was calculated according to the formula h2= rg(mp)2/ (rg(mp)2+re2/r), whererg(mp)2represented the variance component of the genotype nested within the maternal plant, re2 represented the variance component of the error term and r represented the number of replicates (Wricke and Weber1986).
Genotyping
DNA of 254 genotypes was extracted from fresh or fro- zen leaf tissue using the DNeasy 96 plant kit (Qiagen, Hilden, Germany). AFLP analysis was performed as described by Herrmann et al. (2005) using 21 primer combinations. The primer combinations were named according to the standard primer combination code (Keygene, Wageningen, Netherlands; see Fig. 1). One hundred and seven SSR primer pairs were screened for polymorphism in the mapping population. SSR assays of primer pairs selected from TPSSR01 to TPSSR57 (Ko¨lliker et al.2005) were performed using the protocol of Ko¨lliker et al. (2005) and 5¢ pigtail addition to the reverse primer to promote non-templated adenylation of amplicons (Brownstein et al.1996). Forward primers of SSR loci RCS004–RCS233 reported by Sato et al. (sub- mitted) were modified by 5¢concatenation of the M13-18- mer 5¢-TGTAAAACGACGGCCAGT-3¢, which per- mitted concurrent fluorescence labelling of PCR prod- ucts by a third primer (M13) with an incorporated fluorophore (Boutin-Ganache et al. 2001) and reverse primers carried the 5¢ pigtail. PCR analyses were con- ducted in a total volume of 20 ll containing 15 ng DNA, 1· PCR buffer, 0.1lM of forward primer, 0.4lM of M13 and reverse primer, 2.5 mM MgCl2, 0.25 mM of each dNTP and 0.5 UTaqDNA Polymerase (Invitrogen, Carlsbad, CA, USA). PCR conditions consisted of 4 min at 95C, 30 cycles of 30 s at 95C, 30 s at 50C or 55C depending on the primer pair, 30 s at 72C, 10 cycles of 30 s at 95C, 30 s at 53C, 30 s at 72C followed by a final extension of 10 min at 72C.
AFLP and SSR amplification products were analysed on an ABI Prism 3100 genetic analyzer using POP-4 polymer and 36 cm capillaries, and scored using the Genotyper 3.6 software (Applied Biosystems, Foster City, CA, USA).
Linkage mapping and QTL analysis
A genetic linkage map was established using JoinMap (version 3.0; Van Ooijen and Voorrips2001) and the CP (cross-pollination) population type. For grouping, a LOD threshold of four or lower was obtained. The order of loci was determined at LOD larger than 1.0, REC smaller than 0.4 and a jump threshold of five using Kosambi’s mapping function.
QTL analysis was performed based on least square means of the genotypes using MapQTL (version 5.0, Van Ooijen 2004). In order to reduce computing time necessary for calculations with the CP population type and the MQM (multiple QTL model) algorithm, the number of loci was optimised for each trait as follows:
In a first step two maps were calculated for each parental plant, i.e. one map was calculated with loci heterozygous in pV, the other map contained loci het- erozygous in pC. QTL analysis was performed on these maps for all traits using composite interval mapping (CIM) of PlabQTL with the genotypes coded as a doubled haploid (DH) population type (Utz and Mel- chinger2003). In a second step, loci near putative QTLs identified in step one on both maps, as well as most bi- parental SSR loci were included to construct one com- bined reduced map with an average locus distance of approximately 10 cM. QTL analysis was then performed on this reduced map for all eight traits using backward cofactor selection and MQM mapping. In a last step, separate maps were constructed for each trait with in- creased locus density in regions where putative QTLs were identified in step two. Final QTL mapping was performed with backward cofactor selection and MQM mapping using these maps optimised for each trait.
QTLs were taken into account when the LOD score was higher than the LOD threshold derived from the respective map using 10,000 permutations.
Results
Establishment of the mapping population and phenotypic evaluation
Of the 200 seeds harvested from each parental plant, 176 and 172 produced viable genotypes from pV and pC, respectively. Of these, 82% (140 from pV and 145 pC) yielded at least four equivalent clonal replicates. To in- clude the same number of genotypes from each maternal plant, the final mapping population consisted of 280 F1
genotypes.
Analyses of variance of the phenotypic data revealed highly significant variation among the 280 genotypes for all eight seed yield components studied (Table1). No significant effect of the maternal plants was observed, i.e.
there was no difference between genotypes where seed was harvested from pV when compared to genotypes where seed was harvested from pC. The proportion of variance explained was highest (0.71) for TSW followed
by SNH (0.65) and TOF (0.64; Table1). Heritability ranged from 0.51 for SYP to 0.85 for TSW (Table2), whereas coefficients of variation ranged from 0.06 for TOF to 0.33 for SNP.
Pairwise comparison of SYP with the other seven seed yield components using product moment correla- tion revealed most of them to be highly significantly (P<0.01) correlated with the exception of TOF which
0
10
20
30
70 40
60 50
80 cM
5
C_RCS0007 C_E41/M59_127 C_E39/M59_265 V_P38/M15_126 V_E41/M59_130
V_E32/M48_190* C_P35/M15_164 B_RCS0060 B_P39/M16_277 B_E35/M48_286
V_P38/M15_325
C_P38/M15_332 C_P38/M17_88 C_P35/M15_214 C_E39/M59_380 C_RCS0131 C_P35/M18_142 C_TPSSR17 C_P39/M16_293 C_P32/M15_311 V_P32/M15_132 B_E38/M59_61 V_E39/M48_164 B_E41/M59_201 C_P35/M15_100 C_E35/M48_122 B_E35/M48_126 V_E35/M48_123 C_E39/M48_161 B_P41/M18_62 C_E41/M48_91 V_E39/M59_223 V_E35/M48_77 V_E38/M59_301 B_E39/M50_177 C_E39/M48_151 C_E32/M48_136 C_P32/M15_298 SNHTSWTOF TOF
70 cM
2
C_P42/M15_284 C_TPSSR52 C_E32/M48_55 V_P35/M15_96 V_P42/M15_289 C_E39/M48_122
V_TPSSR15 C_P35/M18_236
C_E38/M59_63
V_E39/M59_161 B_RCS0075 B_RCS0078 B_P41/M18_103
B_E39/M59_150 B_P35/M18_281 C_E39/M48_171 V_E39/M48_77 V_P41/M15_177 V_E39/M48_208 V_E39/M50_94 V_P35/M15_308 V_P38/M15_309 V_E39/M48_194 C_P41/M15_109 C_RCS0167 V_P32/M15_65 B_TPSSR44 C_E39/M50_193 B_RCS0071 V_P39/M16_268 B_TPSSR24 C_P32/M18_208 B_TPSSR05 B_RCS0010 C_P32/M18_365 C_P38/M17_302 C_RCS0074 C_E39/M59_374 V_E42/M50_161 V_P38/M17_298 V_E41/M59_121 C_E38/M59_88 V_E41/M48_208 V_E41/M48_250 V_E39/M50_97 C_P41/M18_83 C_P39/M16_147 V_E35/M48_188 V_P35/M15_155 C_RCS0025 V_RCS0130
TSW
SYH TOFSNHTOF
SYH
78 cM
4
C_P41/M15_144 C_P42/M15_122 C_E39/M50_110 B_E32/M48_263 V_E39/M48_265 V_E41/M59_144 C_P32/M15_319 C_E41/M48_201 B_E39/M59_122 V_E39/M50_288 B_E41/M48_218 B_E41/M59_215
C_E38/M59_230 C_P35/M18_147 C_E39/M59_154
V_E35/M48_322 V_P35/M18_148
V_P39/M16_109 C_E38/M59_74 C_P41/M15_236 C_E32/M48_284 C_P42/M15_327 C_TPSSR45 V_E39/M59_140
V_E35/M48_76 C_E41/M48_161
V_P41/M18_280 V_P32/M18_333 C_P38/M18_237 V_E41/M48_157 C_E41/M48_155 B_E38/M59_333 C_E38/M59_331 C_E42/M50_154 B_RCS0008 C_E39/M48_73 V_RCS0233 B_E42/M50_275 C_P32/M15_289 V_E39/M59_104
SYP
SNP HNPTOF
67 cM
7
V_RCS0132 V_E38/M59_277 C_P35/M15_183* C_E39/M59_178* V_P41/M16_221 C_E35/M48_223**
B_RCS0004**
V_P41/M16_161 B_E41/M48_173 B_E41/M48_166 V_P38/M18_302 V_E39/M50_318* V_P35/M15_182* B_RCS0051**
C_E39/M59_306**
B_E39/M50_146* B_RCS0011**
C_E39/M50_121**
B_P39/M16_326 C_E35/M48_152**
B_TPSSR10*
C_P32/M18_69**
C_P38/M17_180 B_P38/M17_184 C_E39/M48_175* C_E39/M48_304 V_P38/M17_179 B_E39/M59_269 C_E35/M48_149 B_P41/M16_201 C_E35/M48_183 B_P41/M18_198 TSWPSSSNHTOF
54 cM
1
C_P41/M16_71 C_E41/M59_288 C_E41/M59_314 C_E38/M59_224 V_E38/M59_174 B_RCS0153 B_E41/M59_181 V_P38/M15_282
V_E35/M48_244 C_TPSSR40 C_RCS0005 V_P39/M16_346 V_P38/M17_132 V_P38/M15_151 B_TPSSR56 C_E42/M50_148 C_E35/M48_171 C_P38/M15_162 B_TPSSR50 C_P32/M18_218 B_RCS0035 C_E41/M48_232 V_E38/M59_151* C_E39/M48_60
V_E39/M59_256* V_P42/M15_109 C_P39/M16_118 V_P38/M15_119 C_P41/M18_163 C_P39/M16_173 V_E35/M48_375 C_P42/M15_312 V_E39/M50_255 C_P32/M18_258 V_P32/M18_256 V_RCS0198 V_P42/M15_226 PSSHNPTSWSNH
60 cM
6
C_P35/M15_248 C_P32/M15_175 C_P38/M18_273 C_TPSSR54 C_P41/M16_74 B_P41/M16_210 C_E41/M59_81 V_E41/M59_79 B_P42/M15_140 C_E38/M59_210 V_P35/M15_327*
C_E42/M50_361***
B_P35/M18_274* C_P35/M18_278 B_P35/M15_131**
V_P42/M15_174***
V_E39/M48_149****
B_TPSSR23***
C_E39/M59_209 C_P39/M16_342 V_E39/M59_232***
V_E39/M59_291***
B_TPSSR28**
V_E39/M59_201***
V_RCS0031**
SNH SYH HNPTOF
SYH SYP SNP
PSS SNH 54 cM
3
B_E35/M48_316 B_P32/M18_107 C_E35/M48_174 B_P32/M18_137 B_TPSSR09 V_E41/M59_156 C_P32/M15_252* B_TPSSR46 V_P35/M18_199 C_E42/M50_284* V_E35/M48_98 V_P35/M18_220 C_E35/M48_282 C_P32/M15_340 V_RCS0193 V_E39/M48_197 V_E42/M50_146 B_P39/M16_195 B_P39/M16_197 B_RCS0047 C_P32/M15_184 C_E39/M59_90 V_E38/M59_360 C_P35/M18_229 B_E41/M59_107 C_E42/M50_168 C_P38/M18_83 V_P38/M15_87 V_E41/M59_71 V_P39/M16_226 V_P42/M15_105
B_TPSSR16 V_E39/M48_285 C_RCS0169 C_E42/M50_166 SNHTOF TSW
SNH HNP SNP SYP
61 cM
Fig. 1 Genetic linkage map of a red clover population based on 254 F1genotypes, 42 SSR and 216 AFLP loci. Locus names consist of a denomination of the origin of the parental alleles (B = bi- parental locus; C and V = mono-parental locus heterozygous in the parent from the cultivar Corvus and Violetta, respectively), followed by the locus name (standard primer combination code (Keygene, Wageningen, Netherlands) followed by the allele size in relative migration units for AFLP loci or the prefix TPSSR (Ko¨lliker et al.2005) and RCS (Sato et al., submitted) followed by an identification number for SSR loci, respectively). Significantly distorted loci are indicated by asterisks (*P£0.05, **P£0.01,
***P£0.001, ****P£0.0001). Positions of QTLs for eight seed yield components were calculated using MQM mapping, the optimised map for the respective trait and least square means of four replicates per genotype (SYPseed yield per plant;SNPseed number per plant;SYHseed yield per head;SNHseed number per head;HNPhead number per plant; TSWthousand seed weight;
PSSpercent seed set;TOFtime of flowering). The maximum LOD score position of each QTL is indicated with anarrowand abar represents the interval between two positions obtained at LOD scores two units lower than the maximal score
Table 1 Fvalues, level of significance and proportion of variance explained (R2) of analysis of variance for eight seed yield components of a red clover population consisting of 280 F1genotypes assessed in a field experiment with four clonal replicates
df SYP SNP SYH SNH HNP TSW PSS TOF
Maternal planta 1 0.7NS 1.9NS 1.0NS 2.2NS 0.7NS 1.6NS 1.0NS 0.4NS
Genotype (maternal plant) 278 1.9*** 2.3*** 3.1*** 4.4*** 2.2*** 5.9*** 3.1*** 4.8***
Column 3 18.5*** 18.0*** 19.1*** 23.0*** 10.7*** 4.8** 11.1*** 5.5***
Row 3 13.0*** 9.7*** 26.7*** 10.8*** 9.3*** 43.6*** 3.2* 1.0NS
Error 834
R2 0.47 0.50 0.58 0.65 0.48 0.71 0.56 0.64
SYPseed yield per plant;SNPseed number per plant;SYHseed yield per head;SNHseed number per head;HNPhead number per plant;
TSWthousand seed weight;PSSpercent seed set;TOFtime of flowering
aThe population was based on reciprocal crosses where the parent from the cultivar Violetta (pV) served as maternal plant for one half of the genotypes and pC for the other half
*P£0.05,**P£0.01,***P£0.001,NSnot significant
showed only moderately significant (P<0.05) correla- tion and TSW which was not significantly correlated to SYP (Table3). Thereby, the correlation coefficients were highest for comparisons of SYP, SNP and HNP. Pair- wise comparisons among the other seven seed yield components revealed two-thirds of them to be signifi- cantly (P<0.05) correlated. For example, TSW and PSS as well as HNP and TOF showed a significantly negative correlation coefficient of 0.28 and 0.26, respectively (Table3).
Linkage mapping
Forty-two (39%) of the 107 SSR loci analysed yielded polymorphisms among the 254 F1 genotypes used for mapping (Table4). AFLP analysis yielded a total of 216 polymorphic loci. The nine AFLP primer combinations based on the restriction enzymes EcoRI/MseI extended by three selective nucleotides (Eco+3/Mse+3) yielded 5–19 polymorphic loci each, with a total of 117. The 12 Pst+3/Mse+2 primer combinations yielded 4–13 with a total of 99 loci (Table 4).
All 258 loci were mapped onto seven linkage groups (LG; Fig. 1). The length of the LGs ranged from 54 cM (LG 7) to 78 cM (LG 2) with an average of 63.5 cM, resulting in a total map length of 444.2 cM (Fig.1).
Average distance between loci varied from 1.5 (LG 2) to 2.2 (LG 6) with an average of 1.7 cM across the entire map.
Of the 258 loci 58 (22.5%) were bi-parental, i.e. both parents were heterozygous for that locus, whereas the remaining 200 (77.5%) loci were mono-parental, i.e.
heterozygous in one parent and homozygous in the other parent (Table 4). No gap larger than 10 cM be- tween two loci across the linkage map and no clustering of loci depending on the marker system or the restriction enzymes used was observed (Fig. 1).
The percentage of distorted segregation (P£ 0.05) ranged from 5% for the 19 mono-parental SSR loci to 26% for the 23 bi-parental SSR loci resulting in an average of 12% for the total of 258 loci (Table4).
Highly distorted loci (P£0.001) were mainly observed on LG 6 where 48% of the loci were distorted. On LG 7, 47% of the loci were distorted but distortion was less severe (P>0.001; Fig.1).
QTL analysis
A total of 38 significant QTLs, with a LOD score higher than the LOD threshold calculated from the respective trait and the corresponding map, were detected across all eight seed yield components (Table5; Fig.1). Three to eight QTLs were found per trait explaining together 33.8–69.1% of the total variance with an average of 48%
across all traits. Individual QTLs explained 2.1–33.7%
of the variance with an average of 10%. Two QTLs explained more than 30% (TSW and PSS), two ex- plained 20–30% (SNH and TOF), whereas eight QTLs explained 10–20% of the total variance (Table5). Seven additional QTLs were identified with LOD scores only slightly lower than the LOD threshold, i.e. two QTLs for SYP (LG 5, 51 cM; LG 1, 44 cM), two for SNP (LG 5, 37 cM; LG 1, 45 cM), one for TSW (LG 6, 48 cM) and two for PSS (LG 2, 78 cM; LG 1, 0 cM; data not shown).
On each LG four to nine significant QTLs were de- tected and up to five QTLs for different traits were lo- cated within 10 cM (LG 6, 44–54 cM). QTLs of highly correlated traits (Table3) were often detected within 1–
10 cM (Fig.1). For example, all QTLs for SYP and for SNP, traits which showed a correlation coefficient of 0.95, were identified within 1 cM (LGs 3 and 6) and 6 cM (LG 4), respectively. In addition, of the seven QTLs identified for SYP and HNP, traits which showed a correlation coefficient of 0.87, two sets of two QTLs were located within 6 cM (LGs 3 and 4). On the other hand, QTLs of insignificantly correlated traits (Table3) were often located on separate LGs or at least were not identified in the same region (Fig.1). For example six QTLs of SYP and TSW, traits which showed an insig- nificant correlation coefficient of only 0.08, were de- tected on separate LGs, whereas two QTLs were located 37 cM apart on LG 3 (Fig.1).
Discussion
The explanatory power of trait dissection and QTL analysis largely depends on a reliable assessment of the phenotypic variation for the traits under study. In out- crossing species, where homozygous lines are difficult to obtain, replicated field experiments often rely on clonal replicates of individual genotypes. However, for red clover such replicates are difficult to obtain (Cumming and Steppler 1961), because the tap root system is less suitable for cloning than the fibrous root system of grasses or the stoloniferous growth of white clover.
Establishing plants under short-day conditions to enhance shoot formation and dipping cutlets in synthetic
Table 2 Key characteristics of eight seed yield components of a red clover population consisting of 280 F1genotypes assessed in a field experiment based on least square means of four replicates per genotype
Mean Minimum Maximum SDa h2 b
SYPc 10.0 0.9 19.3 3.1 0.51
SNP 5,722 713 11,309 1,870 0.58
SYH 0.14 0.07 0.21 0.02 0.70
SNH 79 48 110 11 0.79
HNP 72 12 140 21 0.57
TSW 1.75 1.27 2.21 0.15 0.85
PSS 0.76 0.51 0.93 0.08 0.70
TOF 49 38 62 3 0.80
aStandard deviation
bHeritability
cFor description of seed yield components, see Table1
auxin to enhance root formation was highly successful with 82% of the genotypes producing four or more clonal replicates. This method provides a valuable pre- requisite for further QTL analysis in red clover.
Besides reliable phenotypic data, a robust linkage map with evenly distributed markers is needed. The linkage map developed here consisted of seven linkage groups and was based on 258 SSR and AFLP loci (Fig.1). As the map in this study and the map reported by Sato et al. (submitted) included common SSR loci, and the map reported by Isobe et al. (2003) in turn in- cluded RFLP loci present on the map of Sato et al.
(submitted), a congruent numbering of LGs was used for all the three maps. The total map with a length of 444.2 cM was only slightly shorter than the 535.7 cM of the red clover map based on 157 RFLP loci reported by Isobe et al. (2003), but exhibited only half the length of a very recently developed map based on 1,505 SSR and RFLP loci (850.4 cM; Sato et al., submitted). The large difference in locus number may be one reason for the different map lengths, as was also reported for other species, such as perennial ryegrass (Armstead et al.2002;
Bert et al.1999; Jones et al.2002). Other factors, such as the heterogeneity of the parents, i.e. the particular parental genotype, may also play an important role.
Since longer maps only rarely covered additional gen- ome regions, when compared to shorter maps of the same species (Cregan et al.1999a; Freyre et al.1998), the 258 loci used in this study are likely to cover the majority of the red clover genome and to be sufficient for QTL
analysis. However, complete genome coverage of the map presented here cannot be assumed.
For enhanced accuracy and power of QTL detection, increased number of genotypes rather than the number of loci used for analysis are crucial. A population size of at least 200 genotypes is needed to detect a QTL with an explained variance of 5% (Van Ooijen 1992). On the other hand, the power of detecting a QTL remains vir- tually the same no matter a map with an average locus distance of 10 cM or with an infinite number of loci is used (Darvasi et al.1993). The average locus distance of 1.7 cM obtained in the present map therefore provided a good basis for QTL analysis.
In addition, the two marker systems used in this study complemented one another and allowed the construc- tion of a meaningful map. On the one hand, the inte- grated SSR loci can be used to link the map with the other published red clover maps (Isobe et al.2003; Sato et al., submitted) providing a valuable basis for further investigation of genome regions of interest. On the other hand, the AFLP loci offer a powerful tool to quickly fill gaps between SSR loci on linkage maps. One drawback of the AFLP marker system might be the clustering of loci in centromeric regions observed using the restriction enzyme EcoRI (Bert et al. 1999; Vuylsteke et al.1999).
Thus, to avoid clustering of AFLP loci but still ensuring coverage of centromeric regions, both rare cutting restriction enzymes, EcoRI and PstI, were used. Al- though, the average distance between two loci of the map was comparable to maps in which clustering of
Table 3 Product moment correlation coefficients for pairwise comparisons of eight seed yield components of a red clover population consisting of 280 F1genotypes
SYPa SNP SYH SNH HNP TSW PSS
SNP 0.95***
SYH 0.42*** 0.32***
SNH 0.42*** 0.47*** 0.79***
HNP 0.87*** 0.89*** 0.02NS 0.10NS
TSW 0.08NS 0.19** 0.39*** 0.20** 0.09NS
PSS 0.22*** 0.28*** 0.31*** 0.47*** 0.09NS 0.28***
TOF 0.16* 0.23*** 0.11NS 0.03NS 0.26*** 0.15* 0.11NS
*P<0.05,**P<0.01,***P<0.001,NSnot significant
aFor description of seed yield components, see Table1
Table 4 Number of AFLP and SSR loci of a red clover linkage map based on 254 F1genotypes
Bi-parental locia Mono-parental locib Total
abxcdc efxeg hkxhk lmxll (pC) nnxnp (pV)
SSR 13 (23%) 10 (30%) – 12 (0%) 7 (14%) 42 (17%)
AFLPEco/Msed – – 19 (5%) 52 (15%) 46 (17%) 117 (15%)
AFLPPst/Msed – – 16 (13%) 47 (6%) 36 (8%) 99 (8%)
Total 13 (23%) 10 (30%) 35 (9%) 111 (10%) 89 (13%) 258 (12%)
The percentage of distorted loci is indicated in parentheses
aBoth parents were heterozygous for these loci
bHeterozygous in one parent [from the cultivar Corvus (pC) or Violetta (pV)] and homozygous in the other parent
cSegregation types according to JoinMap (Van Ooijen and Voorrips2001)
dAFLP primer combinations based on the restriction enzymesEcoRI andMseI orPstI andMseI, respectively
EcoRI loci was observed (Bert et al.1999), we observed no such clustering (Fig. 1).
Due to the manifestation of deleterious recessives, inbreeding depression may lead to substantial segrega- tion distortion and may negatively influence the stability of genetic maps as well as the accuracy of QTL analyses (Brummer et al.1993; Echt et al.1994). The use of non- inbred F1populations may be an effective way to avoid distortion in self-incompatible species, such as red clover (Tavoletti et al.1996). Indeed, the proportion of signif- icantly (P £0.05) distorted loci observed (Table4) was only 12%, which is considerably lower than the 37%
observed in a backcross population of red clover (Isobe et al.2003). Self-incompatibility may be another reason for segregation distortion and may result in a clustering of distorted loci around a self-incompatibility locus (Bert et al. 1999). Thus, the region on LG 6, where highly distorted loci were observed, might correspond to the single self-incompatibility locus reported in red clo- ver (Lawrence 1996). Segregation distortion based on self-incompatibility can only occur if the parents share a common self-incompatibility allele. If this were the case, not all of the three resulting genotypes would have the same probability to be successfully pollinated. If pollen
Table 5 Position and description of QTLs identified using MQM mapping, the optimised map for the respective trait and least square means of eight seed yield components of a red clover population consisting of 254 F1genotypes
Trait Linkage group Position (cM) Closest neighbouring locus Maximum LOD scorea Percentage variance explained
SYPb 3 0.6 C_E35/M48_174c 10.37 15.3
4 67.3 V_E39/M59_104 4.90 7.0
6 45.3 V_E39/M59_291 7.60 11.5
Total 33.8
SNP 3 1.6 B_P32/M18_137 9.89 14.0
4 61.4 V_RCS0233 6.02 8.1
6 44.3 V_E39/M59_291 9.65 14.4
Total 36.5
SYH 2 9.8 V_TPSSR15 6.56 8.1
2 47.6 B_TPSSR24 7.60 6.4
6 3.2 C_P32/M15_175 3.88 7.7
6 47.3 B_TPSSR28 11.05 17.5
Total 39.7
SNH 1 2.1 C_E41/M59_288 8.86 8.2
2 6.2 C_E39/M48_122 4.79 2.6
3 1.6 B_P32/M18_137 3.94 3.4
3 44.9 V_P42/M15_105 5.49 5.4
5 36.6 C_P35/M15_214 5.04 2.1
6 3.2 C_P32/M15_175 4.81 7.0
6 54.2 V_RCS0031 21.74 25.7
7 48.3 C_E35/M48_149 11.25 8.5
Total 62.9
HNP 1 43.9 V_E39/M59_256 5.35 6.7
3 1.6 B_P32/M18_137 10.55 13.2
4 61.4 V_RCS0233 7.92 9.9
6 0.0 C_P35/M15_248 4.34 7.7
Total 37.5
TSW 1 35.3 B_RCS0035 5.77 5.5
2 49.6 B_TPSSR05 28.23 32.0
3 38.5 C_P38/M18_83 4.35 4.0
5 11.4 V_E32/M48_190 12.50 12.3
7 20.6 B_RCS0051 5.17 4.7
Total 58.5
PSS 1 51.9 V_P38/M15_119 4.63 5.1
6 54.2 V_RCS0031 18.34 33.7
7 5.5 V_P41/M16_221 3.93 4.7
Total 43.5
TOF 2 0.0 C_P42/M15_284 9.56 8.4
2 44.8 V_P32/M15_65 26.26 23.8
3 49.2 V_P42/M15_105 7.50 6.4
4 56.0 V_E41/M48_157 7.16 5.2
5 37.6 C_E39/M59_380 5.17 4.5
5 58.5 C_E41/M48_91 8.46 7.1
6 35.5 V_P42/M15_174 12.52 10.5
7 43.4 C_E39/M48_304 4.47 3.2
Total 69.1
aSignificant LOD threshold was 3.6 except for SNH where it was 3.7
bFor description of seed yield components, see Table1
cFor description of loci, see Fig.1
is scarce, this may lead to incomplete pollination of florets resulting, first of all, in a negative effect onto PSS.
As a consequence, ghost QTLs for seed yield compo- nents might be observed near the self-incompatibility locus. However, a lack of pollen is very unlikely for the present study since three quarters of the 1,120 individ- uals flowered within 10 days (data not shown) and PSS was comparable to other studies (Oliva et al. 1994).
Moreover, the theoretical proportion of compatible pollen in an isolated progeny of a cross between two parents sharing one self-incompatibility allele varies only between 31.25 and 37.5% for the handicapped and fa- voured genotypes, respectively.
The overall aim of this study was not only to identify QTLs for the development of molecular markers linked to seed yield, but also to elucidate the association among seed yield components. According to the correlation coefficients, SNP and HNP showed the largest effect on SYP (Table3). This result is congruent with path coef- ficient analyses for seed yield in red clover, for which the number of heads was identified as the primary compo- nent affecting seed yield (Montardo et al. 2003; Oliva et al.1994). Some of the seed yield components were not determined independently and showed some mathe- matical causality, partially explaining their correlation.
However, detailed QTL analyses confirmed these asso- ciations since QTLs for SNP, HNP, SNH, SYH and PSS were detected in the same regions (£ 10 cM) like the three QTLs for SYP (Fig.1; Table5). Although all five factors substantially influence SYP, only HNP offers an advantage for phenotypic selection compared to SYP.
HNP is comparatively easy to determine and can be assessed in the field earlier before seed maturity.
Therefore, the selection for increased HNP may present a valuable strategy to improve seed yield in red clover.
On the other hand, TSW showed insignificant cor- relation with HNP and SYP (Table3), which was in congruence with path coefficient analysis in red clover where the influence of thousand-seed weight was minor (Montardo et al.2003; Oliva et al.1994), or with white clover where thousand-seed weight was not correlated with seed yield and inflorescence density (Barrett et al.
2005). Since only two of a total of nine QTLs for HNP and TSW were detected in the same region, successful selection for TSW may be possible independently of the proposed selection for HNP. The QTL located on LG 2 is of particular interest for further investigations to im- prove TSW, as it explains 32% of the variation.
We were able to identify two regions covering less than 10 cM, where five (LG 6) and four (LG 3) QTLs of different seed yield components were clustered, respec- tively (Fig. 1). All but one QTL in these two regions explained more than 10% of the total variation (Ta- ble 5). However, QTL analysis based on segregating populations derived form parents with contrasting phe- notypes has several limitations. The precision and accuracy of QTL detection depends on a large popula- tion size. Small populations lead to an underestimation of the number of QTLs and an overestimation of the
explained variance (Scho¨n et al. 2004) as well as to a limited precision regarding the QTL position (Visscher and Goddard 2004). This is particularly true when assessing traits influenced by a high number of QTLs with small effects (Scho¨n et al. 2004). Moreover, in contrast to more recent approaches like association mapping, only two alleles at a given locus can be studied simultaneously. However, association mapping relies on information about the nature of linkage disequilibrium within the genome of the respective plant species (Flint- Garcia et al. 2003), which is currently not available.
Taken these limitations into account, the presented re- sults may still serve as a valuable base for further molecular dissection of seed yield in red clover. For fu- ture studies, the following approaches may be consid- ered. Fine mapping of the detected QTL, using the information of the recently published high-density SSR map (Sato et al., submitted), offers a promising possi- bility to identify closely linked markers for marker-as- sisted breeding or even the identification of genes involved in the control of seed yield by map-based cloning. The recently published EST resources of the model legume barrel medic (Medicago truncatula; Can- non et al.2005) offer an additional possibility to further explore genetic control of seed yield in red clover. A prerequisite to apply such information is a certain degree of synteny between the target species (i.e. red clover) and the species for which genetic information is available.
The existence of syntenic relationships between a num- ber of legume species including barrel medic, alfalfa (Medicago sativa), soybean, pea and birds foot trefoil (Lotus japonicus) has been demonstrated (Choi et al.
2004). Thus, comparative techniques such as compara- tive anchor marker tag sequences (CATS; Schauser et al.
2005) or single nucleotide polymorphisms (SNP;
Andersen and Lu¨bberstedt 2003) may be alternative approaches for future investigations. Based on genes and QTLs associated with seed yield in other species, such as barrel medic (Cannon et al.2005) or white clover (Barrett et al. 2005), these approaches may help to elu- cidate genetic control of seed yield in red clover. In conclusion, with the stable linkage map obtained using SSR and AFLP loci and 254 genotypes of a F1 popu- lation as well as with the field analysis based on four clonal replicates, a solid basis for QTL analysis for seed yield components was provided. A total of 38 QTLs were detected for the eight seed yield components. The associations among seed yield components allowed the identification of head number per plant as an easy to determine, indirect character to select for seed yield.
Furthermore, two genome regions rich in QTLs for seed yield components were identified with great potential for future characterisation and the development of markers closely linked to seed yield components. To the best of our knowledge, this is the first report on QTL analysis in red clover, which presents an important first step to- wards marker-assisted selection and may help to implement new breeding strategies to complement breeding for complex traits, such as seed yield.
AcknowledgementsThe authors would like to thank Yvonne Ha¨fele for technical assistance in the laboratory, Simone Gu¨nter, Philipp Streckeisen and Peter Tanner for support in the field, Eva Bauer of the State Plant Breeding Institute at the University of Hohenheim, Germany, for assistance with linkage mapping and QTL analysis and Sachiko Isobe of the National Agricultural Research Centre for Hokkaido Region, Japan for information on SSR primer se- quences and map location. This study was funded by the breeding foundation DSP-BLW.
References
Abberton MT, Cogan NOI, Smith KF, Kearney G, Marshall AH, Williams A, Michaelson-Yeates TPT, Bowen C, Jones ES, Vecchies AC, Forster JW (2005) Quantitative trait locus anal- ysis of morphogenetic and developmental traits in an SSR- and AFLP-based genetic map of white clover (Trifolium repensL.).
In: Humphreys MO (ed) Proceedings of the 4th international symposium on the molecular breeding of forage crops and turf, a satellite workshop of the XXth International Grassland Congress, July 2005, Aberystwyth, Wales. Academic, Wagen- ingen, pp 150
Andersen JR, Lu¨bberstedt T (2003) Functional markers in plants.
Trends Plant Sci 8:554–560
Armstead IP, Turner LB, King IP, Cairns AJ, Humphreys MO (2002) Comparison and integration of genetic maps generated from F2 and BC1-type mapping populations in perennial rye- grass. Plant Breed 121:501–507
Barrett BA, Baird IJ, Woodfield DR (2005) A QTL analysis of white clover seed production. Crop Sci 45:1844–1850
Belzile L (1991) The effect of soil type on red clover seed yield. Can J Plant Sci 71:1039–1046
Bert PF, Charmet G, Sourdille P, Hayward MD, Balfourier F (1999) A high-density molecular map for ryegrass (Lolium perenne) using AFLP markers. Theor Appl Genet 99:445–452 Boutin-Ganache I, Raposo M, Raymond M, Deschepper CF
(2001) M13-tailed primers improve the readability and usability of microsatellite analyses performed with two different allele- sizing methods. BioTechniques 31:24–28
Brownstein MJ, Carpten JD, Smith JR (1996) Modulation of non- templated nucleotide addition byTaqDNA polymerase: primer modifications that facilitate genotyping. BioTechniques 20:1004–1010
Brummer EC, Bouton JH, Kochert G (1993) Development of an RFLP map in diploid alfalfa. Theor Appl Genet 86:329–332 Cannon SB, Crow JA, Heuer ML, Wang XH, Cannon EKS, Dwan
C, Lamblin AF, Vasdewani J, Mudge J, Cook A, Gish J, Cheung F et al. (2005) Databases and information integration for the Medicago truncatulagenome and transcriptome. Plant Physiol 138:38–46
Choi HK, Mun JH, Kim DJ, Zhu HY, Baek JM, Mudge J, Roe B, Ellis N, Doyle J, Kiss GB, Young ND, Cook DR (2004) Esti- mating genome conservation between crop and model legume species. Proc Natl Acad Sci USA 101:15289–15294
Cogan N, Smith K, Yamada T, Francki M, Vecchies A, Jones E, Spangenberg G, Forster J (2005) QTL analysis and comparative genomics of herbage quality traits in perennial ryegrass (Lolium perenneL.). Theor Appl Genet 110:364–380
Cregan PB, Jarvik T, Bush AL, Shoemaker RC, Lark KG, Kahler AL, Kaya N, Vantoai TT, Lohnes DG, Chung L, Specht JE (1999a) An integrated genetic linkage map of the soybean genome. Crop Sci 39:1464–1490
Cregan PB, Mudge J, Fickus EW, Danesh D, Denny R, Young ND (1999b) Two simple sequence repeat markers to select for soy- bean cyst nematode resistance conditioned by the rhg1locus.
Theor Appl Genet 99:811–818
Cumming BG, Steppler HA (1961) The control of growth and development in red clover (Trifolium pratenseL.) IV: vegetative propagation and the use of growth regulators. Can J Plant Sci 836–848
Darvasi A, Weinreb A, Minke V, Weller JI, Soller M (1993) Detecting marker-QTL linkage and estimating QTL gene effect and map location using a saturated genetic map. Genetics 134:943–951
Deneufbourg F (2004) Seed yield of forage crop cultivars. Bull Semen 175:20–21
Echt CS, Kidwell KK, Knapp SJ, Osborn TC, Mccoy TJ (1994) Linkage mapping in diploid alfalfa (Medicago sativa). Genome 37:61–71
Flint-Garcia SA, Thornsberry JM, Buckler ES (2003) Structure of linkage disequilibrium in plants. Annu Rev Plant Biol 54:357–
374
Freyre R, Skroch PW, Geffroy V, Adam-Blondon AF, Shirmo- hamadali A, Johnson WC, Llaca V, Nodari RO, Pereira PA, Tsai SM, Tohme J, Dron M et al. (1998) Towards an integrated linkage map of common bean. 4. Development of a core linkage map and alignment of RFLP maps. Theor Appl Genet 97:847–
856
Herrmann D, Boller B, Widmer F, Ko¨lliker R (2005) Optimization of bulked AFLP analysis and its application for exploring diversity of natural and cultivated populations of red clover.
Genome 48:474–486
Isobe S, Klimenko I, Ivashuta S, Gau M, Kozlov NN (2003) First RFLP linkage map of red clover (Trifolium pratenseL.) based on cDNA probes and its transferability to other red clover germplasm. Theor Appl Genet 108:105–112
Jones ES, Mahoney NL, Hayward MD, Armstead IP, Jones JG, Humphreys MO, King IP, Kishida T, Yamada T, Balfourier F, Charmet G, Forster JW (2002) An enhanced molecular marker based genetic map of perennial ryegrass (Lolium perenne) re- veals comparative relationships with other Poaceae genomes.
Genome 45:282–295
Ko¨lliker R, Enkerli J, Widmer F (2005) Characterisation of novel microsatellite loci for red clover (Trifolium pratenseL.) from enriched genomic libraries. Mol Ecol Notes DOI 10.1111/
j.1471–8286.2005.01133.x
Lawrence MJ (1996) Number of incompatibility alleles in clover and other species. Heredity 76:610–615
Lehmann J, Briner HU (1998) Varieties of red clover and meadow fescue in tests. Agrarforschung 5(4):177–180
Mansur LM, Orf JH, Chase K, Jarvik T, Cregan PB, Lark KG (1996) Genetic mapping of agronomic traits using recombinant inbred lines of soybean. Crop Sci 36:1327–1336
Montardo DP, Dall’agnol M, Crusius AF, Paim NR (2003) Path analysis for seed production in red clover (Trifolium pratense L.). Braz J Anim Sci 32:1076–1082
Newbury HJ (2003) Plant molecular breeding. Blackwell, Oxford Oliva RN, Steiner JJ, Young WC (1994) Red clover seed produc-
tion: 2. Plant water status on yield and yield components. Crop Sci 34:184–192
Schauser L, Fredslund J, Heegard Madsen L, Sandal N, Stougaard J (2005) A computational pipeline for the development of comparative anchor tagged sequence (CATS) markers. In:
Humphreys MO (ed) Proceedings of the 4th international symposium on the molecular breeding of forage crops and turf, a satellite workshop of the XXth International Grassland Congress, July 2005, Aberystwyth, Wales. Academic, Wagen- ingen, pp 73–81
Scho¨n CC, Utz HF, Groh S, Truberg B, Openshaw S, Melchinger AE (2004) Quantitative trait locus mapping based on resam- pling in a vast maize testcross experiment and its relevance to quantitative genetics for complex traits. Genetics 167:485–498 Steiner JJ, Smith RR, Alderman SC (1997) Red clover seed pro-
duction: 4. Root rot resistance under forage and seed produc- tion systems. Crop Sci 37:1278–1282
Tavoletti S, Veronesi F, Osborn TC (1996) RFLP linkage map of an alfalfa meiotic mutant based on an F1 population. J Hered 87:167–171
Taylor NL, Quesenberry KH (1996) Red clover science. Kluwer, Dordrecht
Utz HF, Melchinger AE (2003) PLABQTL, version 1.2, a com- puter program to map QTL. Institute of Plant Breeding, Seed
Science, and Population Genetics, University of Hohenheim, Stuttgart
Van Ooijen JW (2004) MapQTL 5, software for the mapping of quantitative trait loci in experimental populations. Kyazma B.V., Wageningen
Van Ooijen JW, Voorrips RE (2001) JoinMap 3.0, software for the calculation of genetic linkage maps. Kyazma B.V, Wageningen Van Ooijen JW (1992) Accuracy of mapping quantitative trait loci
in autogamous species. Theor Appl Genet 84:803–811 Visscher PM, Goddard ME (2004) Prediction of the confidence
interval of quantitative trait loci location. Behav Genet 34:477–
482
Vuylsteke M, Mank R, Antonise R, Bastiaans E, Senior ML, Stuber CW, Melchinger AE, Lubberstedt T, Xia XC, Stam P, Zabeau M, Kuiper M (1999) Two high-density AFLP linkage maps ofZea maysL.: analysis of distribution of AFLP markers.
Theor Appl Genet 99:921–935
White RP, Murray S, Rohweider M (2000) Pilot analysis of global ecosystems: grassland ecosystems. World Resources Institute, Washington
Wricke G, Weber WE (1986) Quantitative genetics and selection in plant breeding. Walter de Gruyter, Berlin
